Genomic Analysis Reveals the Genetic Determinants Associated With Antibiotic Resistance in the Zoonotic Pathogen Campylobacter spp. Distributed Globally

The genus Campylobacter groups 32 Gram-negative bacteria species, several being zoonotic pathogens and a major cause of human gastroenteritis worldwide. Antibiotic resistant Campylobacter is considered by the World Health Organization as a high priority pathogen for research and development of new antibiotics. Genetic elements related to antibiotic resistance in the classical C. coli and C. jejuni species, which infect humans and livestock, have been analyzed in numerous studies, mainly focused on local geographical areas. However, the presence of these resistance determinants in other Campylobacter species, as well as in C. jejuni and C. coli strains distributed globally, remains poorly studied. In this work, we analyzed the occurrence and distribution of antibiotic resistance factors in 237 Campylobacter closed genomes available in NCBI, obtained from isolates collected worldwide, in different dates, from distinct hosts and comprising 22 Campylobacter species. Our data revealed 18 distinct genetic determinants, genes or point mutations in housekeeping genes, associated with resistance to antibiotics from aminoglycosides, β-lactams, fluoroquinolones, lincosamides, macrolides, phenicols or tetracyclines classes, which are differentially distributed among the Campylobacter species tested, on chromosomes or plasmids. Three resistance determinants, the blaOXA–493 and blaOXA–576 genes, putatively related to β-lactams resistance, as well as the lnu(AN2) gene, putatively related to lincosamides resistance, had not been reported in Campylobacter; thus, they represent novel determinants for antibiotic resistance in Campylobacter spp., which expands the insight on the Campylobacter resistome. Interestingly, we found that some of the genetic determinants associated with antibiotic resistance are Campylobacter species-specific; e.g., the blaOXA–493 gene and the T86V mutation in gyrA were found only in the C. lari group, whereas genes associated with aminoglycosides resistance were found only in C. jejuni and C. coli. Additional analyses revealed how are distributed the resistance and multidrug resistance Campylobacter genotypes assessed, with respect to hosts, geographical locations, and collection dates. Thus, our findings further expand the knowledge on the factors that can determine or favor the antibiotic resistance in Campylobacter species distributed globally, which can be useful to choose a suitable antibiotic treatment to control the zoonotic infections by these bacteria.


INTRODUCTION
The discovery and consequent therapeutic use of antibiotics was a remarkable advance in human medicine, which prevented the mortal outcomes of bacterial infections, saving millions of lives during the last century. However, bacteria have evolved through diverse mechanisms, intrinsic and acquired, to withstand the harmful activity of antibiotics. The antibiotic resistance (AR) is mainly generated by the presence of specific resistance genes or point mutations in some housekeeping genes; likewise, the AR can be transferred between bacteria through different mechanisms of DNA exchange, which has greatly increased the occurrence and spread of antibiotic-resistant bacteria worldwide. The development of AR has progressively compromised the effective use of antibiotics, restricting the therapeutic options available to treat the illness produced by antibiotic-resistant pathogenic bacteria. Nowadays, pathogenic bacteria that show resistance to a great diversity of antibiotics represent a serious threat to health worldwide. It has been estimated that infections produced by these bacteria could cause 10 million annual deaths by 2050 (O'Neill, 2014). Faced with this risk to human health, the World Health Organization (WHO) issued a priority global list of antibiotic-resistant pathogenic bacteria for which there is an urgent need to direct research for discovery and development new antibiotics (WHO, 2017). Importantly, in this WHO report, and also in a published analysis from the United States Centers for Disease Control and Prevention (CDC), antibiotic-resistant Campylobacter spp. were cataloged as a serious health hazard in the world (CDC, 2013;WHO, 2017).
In humans, Campylobacter spp. can cause campylobacteriosis, which is considered the leading food-borne zoonosis and the most common cause of gastroenteritis in the world (EFSA and ECDC, 2018) 2 . C. jejuni and C. coli are the Campylobacter species more frequently involved in human gastroenteritis, hence, these two species have been by far the most studied (Man, 2011;Kaakoush et al., 2015). However, other species such as C. concisus, C. lari, C. upsaliensis, and C. ureolyticus, have also begun to be recognized as causative agents of human and animal campylobacteriosis; therefore, they are known as emerging Campylobacter species (Man, 2011).
In 2010, the global burden of Campylobacter infections was 95,613,970 clinical cases; 21,374 deaths and 2,141,926 DALYs (Disability Adjusted Life Years) (Havelaar et al., 2015). Campylobacteriosis may cause mild to severe clinical signs, or even be asymptomatic. The common symptoms of Campylobacter enteric infections include diarrhea (often bloody), fever, abdominal cramps, headache, nausea and/or vomiting. In vulnerable populations, such as very young children, elderly or immunologically compromised patients, this illness can be mortal 2 . Furthermore, other gastrointestinal manifestations or severe life threatening extragastrointestinal complications may appear (Man, 2011;Kaakoush et al., 2015). Because most Campylobacter enteric infections are self-limiting, antibiotic administration is usually not required. Antibiotic therapy is recommended in patients with severe clinical symptoms, relapses, or a prolonged course of infection (Tang et al., 2017). In these cases, fluoroquinolones such as ciprofloxacin, and macrolides such as erythromycin, are the drugs of choice (Ge et al., 2013).
By its importance for public health and food safety, it is necessary to know the resistome of the genus Campylobacter; the resistome is defined as the collection of AR determinants in a specific bacteria or ecological niche (D' Costa et al., 2006;Wright, 2007;Hu et al., 2017). A very high correlation between the genotype and phenotype for AR has been observed in Campylobacter (Zhao et al., 2016;de Vries et al., 2018;Whitehouse et al., 2018). Therefore, the identification of the Campylobacter genotypes associated with AR could help to choose the best antibiotic treatment against infections by Campylobacter species.
The aim of this study was to gain insight into the genetic determinants that constitute the Campylobacter resistome.

Bacterial Genomes
A total of 237 closed genomes (chromosome and plasmid) of Campylobacter spp. were retrieved from the open-access RefSeq: NCBI Reference Sequence Database 3 in March 2019. NCBI accession numbers of the 237 genomes, the information about the host, collection date and geographic location of the strains from which the DNA was extracted, sequenced and annotated, as well as the number of plasmids annotated as an assembly unit, are registered in Supplementary File S1.

In silico Identification of Genes Associated With Antibiotic Resistance
Datasets from publicly available resistance gene databases Comprehensive Antibiotic Resistance Database (CARD 4 ) (Jia et al., 2017) and the National Database of Antibiotic Resistant Organisms (NDARO 5 ), were downloaded (March 2019) and used to identify the presence of genes associated with AR in the 237 Campylobacter genomes, by following two different approaches. In the first one, the BPGA software (Chaudhari et al., 2016) was applied to clustering the resistance genes from CARD and NDARO with those of the Campylobacter genomes, using the USEARCH clustering tool with default parameters (a cutoff set of 50% amino acid identity and 20 random permutations). In the second approach, a BLASTp search was performed 6 with all predicted ORFs from the Campylobacter genomes, against the products of resistance genes from CARD and NDARO, using an E value cutoff of 10E-5, a selected threshold of 50% amino acid identity and minimum coverage of 60% of the query sequence length. Both approaches were compared to ensure the identification of the respective gene associated with AR, on the base of the best hit.

In silico Identification of Point Mutations Associated With Antibiotic Resistance
Specific point mutations known to mediate resistance to fluoroquinolones (gyrA), macrolides (rplD, rplV and 23S rRNA) and streptomycin (rpsL), in C. coli or C. jejuni, were sought in the 237 Campylobacter spp. genomes by using publicly available ResFinder database version 3.2 7 (April 2019) (Zankari et al., 2012). To confirm the point mutations identified by ResFinder, the amino acid sequence of GyrA or RpsL, as well as the nucleotide sequence of 23S rRNA, were aligned and examined using the MEGA X software version 10.1 with the MUSCLE algorithm 8 .
In addition to genes for β-lactamases, we found that 16.5% of analyzed genomes (39/237 genomes) harbor the tet(O) gene (Figure 1), which codes for the ribosomal protection protein involved in resistance to tetracyclines (Sougakoff et al., 1987). Consistent with previous reports indicating the presence of tet(O) either on plasmid or chromosome (Pratt and Korolik, 2005;Dasti et al., 2007), we found that 30 and 9 of the analyzed genomes carry this resistance gene on plasmid and chromosome, respectively ( Table 1). It should be noted that Campylobacter plasmids are usually classified according to the genes they carry, those harboring the tet(O) gene have been called pTet plasmids (Marasini et al., 2018). From 87 plasmids annotated as an assembly unit separated from chromosome in the total of Campylobacter genomes tested, 30 were pTet and only one plasmid containing an AR gene other than tet(O) was found; the 31 plasmids harboring AR genes show sizes ranging from 29,115 to 180,543 bp (Figure 2). These pTet plasmids carry the tet(O) gene alone or together with other AR genes; interestingly, all of them, including that no pTet, are present only on genomes from C. coli and C. jejuni (Figure 2).
On another hand, genes conferring resistance to phenicols or lincosamides were detected in a very small proportion of the analyzed genomes. Only 2 genomes harbor the cat gene (prevalence of 0.84%), which codes for a chloramphenicol acetyltransferase that is associated with resistance to chloramphenicol (Wang and Taylor, 1990); in both cases, this gene was located on chromosome (Figure 1 and Table 1).
AR genotypes were identified according to the different combinations of the genetic determinants associated with AR found in the Campylobacter genomes analyzed. A total of 37 AR genotypes were recognized, being the AR genotypes conferring resistance to β-lactams the most predominant FIGURE 2 | Antibiotic resistance patterns from Campylobacter plasmids. Campylobacter species harboring the respective plasmid, NCBI accession number for each plasmid, genes for antibiotic resistance, and size (bp) for each plasmid, are indicated. Distinct antibiotic resistance patterns are displayed in different color.
found. The different AR genotypes and their prevalence on genomes from C. coli, C. jejuni, and other Campylobacter species, are shown in Supplementary Figures S1-S3. bla OXA−61 and tet(O) were the most common AR genotypes present on genomes from C. coli, with a prevalence of 23.8% (5/21 genomes) and 14.3% (3/21 genomes), respectively; whereas bla OXA−184 (prevalence of 41.1%; 58/141 genomes) and bla OXA−61 (prevalence of 28.4%; 40/141 genomes) were the predominant AR genotypes on genomes from C. jejuni; and bla OXA−493 /gyrA T86V and bla OXA−493 were the AR genotypes more common on genomes from other Campylobacter species, with a prevalence of 31.8% (7/22 genomes) and 27.3% (6/22 genomes), respectively (Figure 4). Location (chromosome or plasmid) of the AR determinants for some of the AR genotypes is shown in Figure 3B. All AR genotypes present in the Campylobacter plasmids are displayed in Figure 2.

Species-Specific
Interestingly, our analysis revealed that some genetic determinants for AR are present only in particular Campylobacter species. For instance, despite the great representativeness of genomes from C. jejuni (163 genomes), and with less abundance from C. coli (22 genomes), the bla OXA−493 gene was only found on genomes from 4 of the 6 species that integrate the C. lari group FIGURE 3 | Resistance or multidrug resistance genotypes harbored on genomes of Campylobacter spp. (A) Multidrug resistance genotype containing the lnu(AN2) gene on a 48,422 bp plasmid from the C. coli RM5611 isolate. (B) Resistance or multidrug resistance genotypes containing different combinations of genetic determinants associated with antibiotic resistance on chromosome and/or plasmid. The name of the Campylobacter species from which the respective genome was obtained, as well as the NCBI accession number and size (bp) of genomes are shown. The colors indicate the antibiotic class to which the respective genetic determinant putatively confers resistance. Brown, β-lactams; pink, tetracyclines; blue, phenicols; green, lincosamides; and red, aminoglycosides. The genes shown with white color are not related to antibiotic resistance. The asterisks indicate truncated genes. The number (n) of genomes harboring each antibiotic resistance genotype is specified on the right. Arrows denote the direction of transcription for each gene. (Miller et al., 2014;Costa and Iraola, 2019): C. insulaenigrae (3/3 genomes), C. lari (7/8 genomes), C. subantarcticus (2/2 genomes) and C. volucris (1/1 genomes) (Supplementary File S1); the number of genomes carrying the bla OXA−493 gene with respect to the total of genomes tested for each species, is indicated between parenthesis. Hence, bla OXA−493 was the predominant gene for β-lactamases that was identified in the C. lari group (prevalence of 92.9%, 13/14 genomes); only one genome from this group presents the bla OXA−184 gene and none carries the bla OXA−61 or bla OXA−576 genes. In contrast, the bla OXA−61 gene was confined to C. jejuni (63/163 genomes) and C. coli (14/22 genomes) species, as well as the bla OXA−184 gene (66 genomes) was mostly present in C. jejuni (65/163 genomes); only one genome of C. helveticus also carries the bla OXA−184 gene (1/1 genomes) (Supplementary File S1). Furthermore, aminoglycoside resistance genes were only detected in C. coli (12/22 genomes) and C. jejuni (11/163 genomes) (Supplementary File S1).

Prevalence of AR Genotypes in Campylobacter and Its Association With Collection Date, Host and Geographic Location
Genetic factors associated with AR (genes or point mutations) were identified in 77.6% of the Campylobacter genomes analyzed (184/237 genomes; 1 to 8 factors per genome). No genetic determinants associated with AR were found on genomes from FIGURE 4 | Frequency of the most prevalent antibiotic resistance genotypes on genomes from C. coli, C. jejuni and other Campylobacter species. Other Campylobacter species include: C. avium, C. fetus, C. helveticus, C. hyointestinalis, C. iguaniorum, C. insulaenigrae, C. lanienae, C. lari, C. peloridis, C. sputorum, C. subantarcticus, C. ureolyticus, and C. volucris. For each genotype, the antibiotic class to which they putatively confer resistance is indicated.
C. concisus, C. cuniculorum, C. curvus, C. gracilis, C. hepaticus, C. hominis, and C. pinnipediorum. Genomes carrying factors that putatively confer resistance to 1 or 2 classes of antibiotics were considered to represent a resistance genotype, those carrying factors related with resistance to 3 or more antibiotic classes were considered to be a multidrug resistance (MDR) genotype, and those lacking factors for AR were taken as susceptible genotypes. Most studies on AR in Campylobacter have been performed on the C. coli and C. jejuni species, whereas this phenomenon remains poorly explored for the other Campylobacter species. Thus, to gain more insight about the AR in C. coli and C. jejuni, as well as in the Campylobacter species less studied, we decided to compare the prevalence of resistance or MDR genotypes between genomes from C. coli (22 genomes), C. jejuni (163 genomes) and the other Campylobacter species analyzed (52 genomes). For the genomes from C. coli, the resistance genotype is the most represented (prevalence of 63.6%; 14/22 genomes), followed by the MDR genotype (prevalence of 31.8%; 7/22 genomes) and finally the susceptible genotype (prevalence of 4.6%; 1/22 genomes) (Figure 5). For genomes from C. jejuni, the resistance genotype is also the predominant (prevalence of 78.5%; 128/163 genomes), then the susceptible genotype (prevalence of 13.5%, 22/163 genomes) and at last the MDR genotype (prevalence of 8%; 13/163 genomes) (Figure 5). In contrast, for the genomes from the other Campylobacter species, the susceptible genotype is predominant (prevalence of 57.7%; 30/52 genomes), followed by the resistance genotype (prevalence of 40.4%; 21/52 genomes) and only one MDR genotype was found (prevalence of 1.9%; 1/52 genomes) (Figure 5).
Then, we analyzed the prevalence of the Campylobacter AR genotypes according to the host, collection date and geographical location from where the respective bacteria were isolated (Supplementary File S1). The C. coli isolates from farm animals carry resistance or MDR genotypes, with prevalence of 61.1% (11/18 genomes) and 33.3% (6/18 genomes), respectively, whereas the only 2 C. coli isolates from humans present the resistance genotype ( Figure 6A). For C. jejuni, the MDR genotype has a higher prevalence in isolates from farm animals (25.8%; 8/31 genomes) compared with those from humans (3.7%; 4/109 genomes); the opposite was observed for the resistance genotype ( Figure 6B). For the other Campylobacter species, the resistance genotype is predominant in bacteria isolated from marine animals (58.3%; 7/12 genomes), whereas the only MDR genotype found in this group is present in the C. lanienae NCTC 13004 strain isolated from a human (6.7%; 1/15 genomes) ( Figure 6C). Notably, our analyses show a considerable increase in the presence of MDR genotypes from C. coli ( Figure 7A) and C. jejuni (Figure 7B), as well as in the presence of resistance genotypes from the other Campylobacter species group (Figure 7C), in bacteria collected in the last decade (2010-2018), compared with previous decades. Additionally, our results indicate a very high prevalence of AR genotypes in C. coli and C. jejuni isolates from both Europe and North America (88-100%); being the MDR genotype from both C. coli and C. jejuni more predominant in North America (Figures 8A,B). Likewise, our analyses indicate a higher prevalence of AR genotypes from the other Campylobacter species group in Europe (56.3%) with respect to North America (35.7%) (Figure 8C).

DISCUSSION
Previous studies have analyzed the prevalence of genetic determinants for AR in genomes from C. coli and C. jejuni FIGURE 5 | Frequency of susceptible, resistance or multidrug resistance (MDR) genotypes on the 237 Campylobacter genomes tested. Frequencies for genomes from each C. coli, C. jejuni and other Campylobacter species (C. avium, C. concisus, C. cuniculorum, C. curvus, C. fetus, C. gracilis, C. helveticus, C. hepaticus, C. hominis, C. hyointestinalis, C. iguaniorum, C. insulaenigrae, C. lanienae, C. lari, C. peloridis, C. pinnipediorum, C. sputorum, C. subantarcticus, C. ureolyticus, and C. volucris) are indicated. The genotypes were classified as follows: susceptible, lacks genetic determinants for antibiotic resistance; resistance, carries one or more genetic determinants related to resistance to antibiotics belonging to 1 or 2 different classes; MDR, contains one or more genetic determinants related to resistance to antibiotics belonging to 3 or more different classes. The number (n) of the respective genomes analyzed is shown above the columns. strains isolated in specific geographic regions (Weis et al., 2016;Zhao et al., 2016;Cantero et al., 2018;de Vries et al., 2018;Fabre et al., 2018;Whitehouse et al., 2018). In the present study, we identified genetic determinants associated with AR in 237 genomes from 22 different Campylobacter species with global distribution.
We found 15 distinct AR genes and point mutations related to AR in 3 housekeeping genes in the Campylobacter genomes tested (Figure 1). Notably, 3 of the AR genes had not been previously reported in Campylobacter: lnu(AN2), putatively associated with lincosamides resistance, and 2 putative β-lactams resistance genes, bla OXA−493 and bla OXA−576 . Two β-lactamases coding genes, bla OXA−61 and bla OXA−184 , were the most abundant AR genes in the analyzed genomes from C. coli and C. jejuni, which is consistent with the results from previous studies (Griggs et al., 2009;Weis et al., 2016;Zhao et al., 2016;de Vries et al., 2018;Fabre et al., 2018;Whitehouse et al., 2018). In contrast, in the analyzed genomes from species other than C. coli and C. jejuni, the bla OXA−493 β-lactamase coding gene was the predominant (Supplementary File S1). On another hand, even though our study indicates a low occurrence of determinants associated with aminoglycoside resistance in the Campylobacter genomes tested, these determinants displayed a great diversity, since 9 different aminoglycoside resistance genes, as well as point mutations in rpsL, were found (Figure 1). These findings are in agreement with those reported previously (Qin et al., 2012;Zhao et al., 2015Zhao et al., , 2016Cantero et al., 2018;Fabre et al., 2018;Hormeno et al., 2018;Whitehouse et al., 2018).
A high global incidence of Campylobacter isolates resistant to ciprofloxacin (fluoroquinolones) and erythromycin (macrolides) has been reported before (Luangtongkum et al., 2009;Signorini et al., 2018;Sproston et al., 2018). However, our analysis showed a low occurrence of mutations in the gyrA and 23S rRNA genes (Figure 1), which are associated with resistance to fluoroquinolones and macrolides, respectively. Nevertheless, multidrug efflux pumps can also mediate resistance to antibiotics of different classes, such as aminoglycosides, β-lactams, chloramphenicol, fluoroquinolones, macrolides and tetracyclines, among others (Blair et al., 2014). The efflux pumps are ancient elements that play important physiological roles in bacteria; but notably, they can also export antibiotics out of the cell, thus reducing their intracellular concentration and, as consequence, confer AR (Blair et al., 2014;Blanco et al., 2016). CmeABC is the only multidrug efflux pump characterized in Campylobacter; it was shown to be involved in resistance to structurally different antibiotics including β-lactams, fluoroquinolones, macrolides, and tetracyclines (Lin et al., 2002;Luangtongkum et al., 2009;Guo et al., 2010;Shen et al., 2018). In addition to specific genes and point mutations for AR, we also detected genes that code for multiple efflux pumps in Campylobacter (Supplementary File S2). Genes for the CmeABC system, MacB family, macrolide ABC transporter and TolC family efflux pumps, were found in the bulk of analyzed genomes. Besides, genes for other efflux pumps were found only in some genomes, such as those for the AcrB/AcrD/AcrF family, hydrophobe/amphiphile efflux-1 family and multidrug efflux SMR transport efflux pumps. Important to note, given that the efflux pumps can confer resistance to multiple antibiotics, to facilitate interpretation, they were not considered in our analysis as determinants for AR.
A major finding of our work was the Campylobacter species-specific distribution of some genetic determinants for AR. Campylobacter species cluster in 5 discrete phylogenetic clades (Costa and Iraola, 2019). One is the C. lari group, which is composed of six species: C. insulaenigrae, C. lari, C. ornithocola, C. peloridis, C. subantarcticus, and C. volucris (Miller et al., 2014;Costa and Iraola, 2019). These species are highly related at the genome level and have been isolated from similar hosts and environments (Miller et al., 2014). We found the bla OXA−493 gene and the T86V substitution in GyrA, exclusively in species from the C. lari group (Supplementary File S1). The T86V mutation in GyrA had also been previously reported in the C. lari species (Piddock et al., 2003;Weis et al., 2016). In contrast, aminoglycoside resistance genes were only detected in the classical C. coli and C. jejuni species (Supplementary File S1). Previous studies also have indicated the presence of aminoglycoside resistance genes in FIGURE 6 | Frequency of susceptible, resistance or multidrug resistance (MDR) genotypes on Campylobacter genomes from isolates collected from different hosts. Genotypes from (A) C. coli, (B) C. jejuni and (C) other 20 Campylobacter species: C. avium, C. concisus, C. cuniculorum, C. curvus, C. fetus, C. gracilis, C. helveticus, C. hepaticus, C. hominis, C. hyointestinalis, C. iguaniorum, C. insulaenigrae, C. lanienae, C. lari, C. peloridis, C. pinnipediorum, C. sputorum, C. subantarcticus, C. ureolyticus, and C. volucris. Farm animals include chicken, bovine, pig, sheep, turkey, and rabbit; marine mammals comprise seal, sea lion, penguin, albatross, gull, and shellfish; other animals encompass iguana, lizard, snake, cat, and alpaca. Only those hosts with more available information are displayed. The number (n) of genomes corresponding to each host is shown above the columns. Campylobacter species: C. avium, C. concisus, C. cuniculorum, C. fetus, C. hepaticus, C. iguaniorum, C. lari, C. pinnipediorum, and C. sputorum. Only those decades with more available information are displayed. The number (n) of genomes corresponding to each decade is shown above the columns.
C. jejuni and C. coli (Qin et al., 2012;Zhao et al., 2015Zhao et al., , 2016Cantero et al., 2018;Fabre et al., 2018;Whitehouse et al., 2018); however, the presence of these genes in species other than C. jejuni and C. coli had not been analyzed. Thus, our data show that the genetic variability between the different clades of Campylobacter also involves the determinants for AR. Due to the limited number of genomes from non-C. coli/C. jejuni species (52 genomes), we cannot discard the possibility that aminoglycoside resistance genes are present in these species, but in very low prevalence.
Consistent with previous reports (NARMS, 2012;de Vries et al., 2018;Whitehouse et al., 2018), we found that the C. coli genomes harbor the highest number of AR determinants, compared with the other Campylobacter species analyzed (Supplementary File S1). Furthermore, in agreement with previous studies (Thakur et al., 2010;Wieczorek et al., 2015), our results show a higher prevalence of MDR genotypes in C. coli (31.8%; 7/22 genomes) with respect to C. jejuni (8%; 13/163 genomes). Notably, 86.4% (19/22 genomes) of the C. coli genomes possess 1-4 plasmids; in contrast, only 14.1% (23/163 genomes) of the C. jejuni genomes carry 1 or 2 plasmids, and 25% (13/52 genomes) of the genomes from the remaining 20 Campylobacter species tested contain 1-5 plasmids (Supplementary File S1). Interestingly, most genes associated with AR in C. coli are located on plasmids, whereas in the other Campylobacter species are placed on chromosome (Supplementary File S1), which could explain, at least in part, the higher number of AR determinants and MDR genotypes in C. coli compared with the other Campylobacter species. Remarkably, in non-C. coli/C. jejuni species, AR genes located on plasmids were not identified (Figure 2 and Supplementary File S1), Campylobacter spp. have mechanisms for conjugation and natural transformation, and transferrable AR has been documented in this genus (Taylor and Courvalin, 1988). Moreover, it has been proposed that Campylobacter gained some AR genes from Gram-positive bacteria (Taylor and Courvalin, 1988;Zilhao et al., 1988;Alfredson and Korolik, 2007). Thus, the AR genes located on plasmids, or on other mobile elements, imply a threat for the appearance of new Campylobacter strains resistant to antibiotics.
We found a reduced number of genetic determinants for AR in Campylobacter, in comparison with the huge number of genetic determinants for AR present in other bacteria such as Enterococcus spp. (Torres et al., 2018), Escherichia coli (Poirel et al., 2018) or Salmonella enterica (McDermott et al., 2018), which share hosts and niches with Campylobacter spp. The reason why Campylobacter spp. maintain in general a low number of genetic determinants for AR remains to be determined.
Our results, together with previous studies, reveal a high prevalence of Campylobacter AR genotypes worldwide, not only from the classical C. jeuni and C. coli species, but also, although still with less extent, from emerging Campylobacter species; these AR genotypes can be present in bacteria residing in different hosts such as humans and different animals (Figures 5, 6, 8 and Supplementary File S1). Even more worryingly, our data support that the prevalence of these bacteria carrying AR genotypes has been increasing with time (Figure 7).
Although the phenotype conferred by all the resistance genetic determinants identified in this study needs to be tested, a very high correlation between the presence of genetic determinants for AR and the respective phenotype has been reported in Campylobacter spp., which reaches up to 100% of correspondence for some specific antibiotics (Nirdnoy et al., 2005;Chen et al., 2013;Zhao et al., 2015Zhao et al., , 2016Fabre et al., 2018;Whitehouse et al., 2018). Hence, it has been suggested that analysis of genomic data, for the identification of genetic determinants associated with AR, has the potential to reliably predict resistance phenotypes (Zhao et al., 2016;Whitehouse et al., 2018;Feldgarden et al., 2019).
Thus, our study, together with other reports, provide genetic determinants that can be used to predict AR in Campylobacter spp., which could greatly help to select the best antibiotic therapy against infections caused by these bacteria. Additionally, our study further expands the knowledge on the genetic elements that shape the resistome of the genus Campylobacter and on the scattering of these AR genetic determinants between the Campylobacter species.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
DP-M and VB contributed to the conception and design of the study, as well as the discussion of the results. DR-M, IM-F, RS, and LL carried out the bioinformatics work. DP-M analyzed the bioinformatics results, created the figures, and wrote the manuscript. VB edited the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by a grant from the Consejo Nacional de Ciencia y Tecnología (CONACYT)/México (2017-01-5182) to VB. DR-M was supported by a master fellowship from CONACYT (grant 771346).

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2020.513070/full#supplementary-material FIGURE S1 | Frequency of resistance or multidrug resistance (MDR) genotypes on C. coli genomes. Thirteen antibiotic resistance genotypes were detected in 21 genomes from C. coli isolates. For each genotype, the antibiotic class to which they putatively confer resistance is indicated.
FIGURE S2 | Frequency of resistance or multidrug resistance (MDR) genotypes on C. jejuni genomes. Twenty antibiotic resistance genotypes were detected in 141 genomes from C. jejuni isolates. For each genotype, the antibiotic class to which they putatively confer resistance is indicated.